I always saw his work as something like a guiding principle, not necessarily something that will revolutionize computing as a whole. When you apply traditional computing constraints and boil down some of his ideas of immediacy, seeing the results of inputs quickly, etc, they're not really that earth shattering, but he communicates these ideas with compelling demos that make you think "wow, software can communicate so much more effectively when the interaction isn't painful". But actual details and implementation are left up to the reader.

Years ago at work I was building an ETL tool from the ground up and my coworker and I were watching Bret Victor videos, reading Alan Kay's work, etc, etc. Young and naive. When I built the prototype for the tool, everything happened immediately and the feedback was on the order of milliseconds as you compiled and ran expressions that transformed datasets. That prototype was compelling enough that we based the real product on its ideas, and that whole interaction model carried through into the product. I was always working to make feedback immediate and visual, and absolutely that came at the cost of some additional complexity. But we ended up with a product that people thought was extremely fast (it wasn't really in terms of absolute throughput...but it sure did feel fast) and that people could play around with and shape their data at the same speed that they could think and type, which was pretty cool.

Ultimately adoption among our customers was very good, all things considered (government customers move at their own pace), and I credit a lot of the success of it to the early decisions we made that prioritized joyful rather than hateful interaction.

At the end of the day we were still just pushing pixels around on a screen which is underwhelming I guess, but there were elements of the fantastical Bret Victor demos that shaped how we did things and led to a better outcome.